Vowel reduction and deletion in Apulian and Lucanian dialects with reference to speech rhythm

Researchers in prosody tend to attribute the perception of speech rhythm to alternations between weak and strong syllables. However, acoustic correlates of prominence play a role in listener evaluations (for the Italo-Romance area see Russo 2010, Mairano 2011). Depending on speech rate and listener expectations, local variations in duration, intensity, and pitch can induce different perceptions. This being the case, the unstressed vowel reduction which characterises some of the varieties of southern Italy can be key, as different results may emerge depending on the chosen model and labelling accuracy for deleted or reduced vowels (Schmid 2004, White et al. 2009, Mairano & Romano 2007a-b, Romano et al. 2011, Marotta 2016). This article presents the results of measurements on four dialects from this area. They are analysed within the control/compensation model (Bertinetto & Bertini 2010) as this method softens the effect of the inclusion of a greater number of consonants in consonant intervals resulting from the loss of vowels. In keeping with perception judgements, this model performs well since it succeeds in placing speech samples with deletion into the stress-based area, more so than those varieties with only strong reduction.